[wellylug] Load averages - Gentoo v. Debian

Daniel Pittman daniel at rimspace.net
Tue Mar 23 12:21:50 NZDT 2010


Bret Comstock Waldow <bcw1000 at yahoo.com> writes:
> wellylug-request at lists.wellylug.org.nz wrote:
>>
>> Subject: [wellylug] High load averages but no apparent cause
>> From: David Harrison <david.harrison at stress-free.co.nz>
>> Date: Mon, 22 Mar 2010 16:49:07 +1300
>> To: Wellington Linux Users Group <wellylug at lists.wellylug.org.nz>
>>
>>
>> Hi,
>> Has anyone experienced high-load averages but haven't been able to see
>> processes that are causing it?
>
> The other responses to this are probably on track for what you are asking,
> but I have an observation about this general issue that I will hang off of
> this message, just for general interest.
>
> I'm working through the best set up for my laptop/tablet.  Given various
> requirements, I've implemented the same desktop in Gentoo and Debian Lenny.

[...]

> One thought is that Debian has services running that the Gentoo install
> does not, as I haven't specifically set them up in Gentoo, but they come
> with Debian by default.
>
> I have about 300-500Mbytes more free memory with the Gentoo implementation,
> and the system load looks to be a fairly constant 15% - 30% less, expressed
> as CPU percentage on the same hardware (swapping identical physical disk
> models for each install).

Wow.  Those are some amazing differences!  I would love to know more; we use
Debian here at work, and it is always nice to get a bit more performance back,
so...

Can you tell me which services are run by Debian out-of-the-box that Gentoo
doesn't?  Just the process names would be fine. ;)

I am also curious about the memory: I presume you mean 300-500MB extra cache,
not unused, memory; if it isn't obvious from the list of daemons, can you
identify what uses that?  Even the results from top in memory sort should be
good enough to start from.

Likewise, the CPU use; those are enormous differences, so I would love to know
what processes are actually using the additional CPU.


> Occasionally on the Gentoo implementation, FireFox starts soaking up lots of
> CPU cycles (I currently have about 80 pages opened in tabs).  Perhaps if I
> left it it might relinquish them later, but when I spot it after a few
> minutes (I want to use my tablet) I kill all Firefox sessions and restart
> it, which improves its behaviour after it has finished reloading.  But
> typical behaviour is the noticeably lighter load on Gentoo, where I'm using
> the Firefox package I compile on the machine rather than the binary
> download-and-install-it package.

Again, impressive.  What compilation options are you using for that Firefox
image?  My partner has performance problems like those you describe, and since
they are pretty consistent across platforms for you and her, it would be great
to know which options you are changing to gain back a "noticeable" amount of
performance...

        Daniel
-- 
✣ Daniel Pittman            ✉ daniel at rimspace.net            ☎ +61 401 155 707
               ♽ made with 100 percent post-consumer electrons



More information about the wellylug mailing list